, City University, London
, Atlantic Systems Guild
Pages: pp. 13-15
We wanted this year's IEEE International Conference on Requirements Engineering (RE '04) to be memorable: inspiring keynote presentations on new RE challenges, tutorials on the latest RE techniques, state-of-the-art presentations on popular topics from industry leaders. That sort of thing. And we succeeded—a typhoon and an earthquake measuring 7.4 on the Richter scale ensured that none of the 310 delegates will forget it quickly.
Of course, the program of research papers, industrial experience reports, state-of-the-practice talks, research posters and demonstrations, panels, minitutorials, and keynote speakers was another fine reason to remember RE '04. For the first time, we included industrial-experience reports in the IEEE proceedings to give practitioners a valuable and available resource of good RE practices. State-of-the-practice talks gave practitioners informed overviews of important topics from RE industry leaders. Tutorials on topics from scenarios and creativity to traceability offered practitioners new RE skills. And the program was structured to enable academics and practitioners to break out of the traditional research and practice tracks and share ideas. RE has become an established joint conference—this was the 12th—where practitioners and researchers can meet and explore requirements problems and solutions together.
RE '04's theme, "requirements for design and innovation in a changing world," reflected the increasingly complex technological and project environments in which requirements work is done. Requirements processes are more and more integrated into wider business, system, and product design processes that reflect the development of enterprise-wide solutions, systems of systems, and consumer products.
Predictions by industry watchers such as the Nomura Research Institute that innovation will replace information technology as the new dominant economic force have implications for RE, which must be seen as part of a wider design process. And where better to hold such a conference than in Japan, a country synonymous with technological innovations such as 3G, portable technologies, and consumer electronics that have changed how we live and work. The three theme articles we chose for this issue are based on papers presented at the conference. They, as well as the tutorial and Point/Counterpoint articles in this issue, accurately reflect the conference theme of innovation in RE.
The program reflected the creativity and innovation theme. Nigel Cross from the UK's Open University gave a keynote address on innovative design in domains ranging from building architecture to Formula One racing cars. Based on studies of expert designers, he reported how framing a new problem—a typical RE activity—is intertwined with discovering the essence of a solution to the problem. In many domains, the problem space is too large to explore up front. Expert designers often explore the problem and solution spaces in parallel, using the emerging solution space to decide what information to elicit next about the problem space. They also often look for chinks—that is, omissions and inconsistencies—in problems and requirement specifications that would enable them to discover more innovative solutions. Although the problem-frame concept is familiar to requirements engineers, 1 Cross's work suggests that we must communicate and train people more effectively on the true nature of RE work and implement new processes that explore the interaction between complex problem and solution spaces.
The second keynote also explored the relationship between the problem and solution spaces. Ikuo Minakata from the Matsushita Electric Company, better known as Panasonic outside of Japan, presented his company's vision of how their products will be used. Panasonic envisions a world of networked and ubiquitous computing devices connected through an infrastructure at home and in vehicles, and to government and healthcare providers through electronic services. The company is implementing this vision by developing and integrating solution consumer electronics that, by definition, change the problem space; thus, they pose new challenges for RE. For example, temperature control in computer-controlled homes changes from local human interaction with controller devices to environment sensing and remote human control of the home using other devices such as mobile phones. Such technologies turn traditional requirements techniques on their head. Their innovative use generates new problem spaces that demand new requirements acquisition, specification, and modeling techniques.
In other sessions, we explored the increasing importance of technology-led innovation during RE. In their tutorial on creativity in RE, Suzanne and James Robertson from the Atlantic Systems Guild gave attendees techniques—for example, creativity triggers such as discover the originating business event—which we can use to explore and invent as well as discover and elicit requirements. Later, a motley crew of RE researchers and practitioners gave an adventurous and entertaining minitutorial that described and demonstrated how to invent requirements for new products. A rapid-fire half-hour session led to the outline specification of a new device that would let conference delegates undertake a sightseeing trip in the shortest possible time. Read more about the tension between inventing and eliciting requirements in the Point/Counterpoint on page 48 of this issue.
The conference also explored the role of researchers as innovators in RE and investigated the challenges in transferring RE research results into practice. In the third keynote address, Axel van Lamsweerde from the University of Louvain in Belgium reported lessons learned from 15 years of experience in applying a goal-driven requirements method to real-world projects. One lesson is to keep your methods as simple as possible. Hide your formalisms by using software tools that present heuristics and patterns to requirements engineers. Another lesson is that successful technology transfer requires that you put the right organizational structures in place. van Lamsweerde described a clear separation of roles with respect to academic research, the transfer of results through methods and tools, and their rollout in industrial practices.
The RE '04 conference both confirmed some established RE practices and revealed new trends. Although scenarios and goal-modeling approaches remain pivotal to most requirements processes, the conference revealed—for the first time in our opinion—a convergence toward an agreed model of RE research and practice in the community. Furthermore, practitioners are beginning to share their successful processes and practices.
To support alternative development paradigms, RE must adapt. For example, both the Service-Oriented Requirements Engineering Workshop and Minakata's keynote address identified the need for RE techniques to support the specification, discovery, composition, and delivery of software services made available via the Internet. Similarly, the paper session and state-of-the-practice talk on RE for COTS-based systems revealed the continuing need for better solutions to select and integrate COTS-based systems and product software.
Often when we think about requirements models, we think of formal models that undergo some form of static analysis. RE '04 revealed the emergence of new techniques to visualize and animate requirements models, primarily to validate these models. Some of the reported approaches were beautifully simple but potentially very effective—such as dropping goal models into MS Excel spreadsheets to exploit its graphics features when choosing among solution alternatives. Others were more complex. One paper reported the animation and visualization of labeled transition systems to describe how events of the operational description change the state of basic propositions from which goals are expressed. Another reported the animation of goal-oriented requirements models by automatically generating parallel state machines from goal operationalizations, executing instances of these models, monitoring for property violations, and visualizing concurrent simulations in terms of animated scenes in a domain.
Finally, a panel on RE education revealed an important emerging consensus on how to teach RE theories and practice more effectively. Four panelists reported their experiences and plans for teaching students in universities and training engineers in the workplace. There was honest reporting of past failures as well as more recent successes. A lively audience discussion revealed substantial agreement about more effective solutions for RE education that we hope to explore in next year's RE conference. These included student groups developing software solutions from each others' specifications, encouraging students to learn from making requirements mistakes, using mission-critical systems as case studies, and showing what acceptance test data derived from requirements should look like, so that they know how requirements should be expressed.
The RE '04 conference struck the right balance between consolidation of existing work and the emergence of new RE research challenges and directions. Next year, RE '05 will take place in France. Although Paris doesn't sit on a fault line, we should again expect some earth-shattering results.